Experiments in Web Page Classification for Semantic Web

We address the problem of web page classification within the framework of automatic annotation for Semantic Web. The performance of several classification algorithms is explored on the Four Universities dataset using page text and link information, with a limited-size feature set. Several well-known classification algorithms are evaluated on the task of web-page classification using the text of web pages. When compared to the methods that use link information, the text-based method shows surprisingly good performance, even with a feature set of limited size.